We’re serious news junkies here at Nanalyze, with a heavy lean toward anything tech-related, which means our knowledge of pop culture and politics stops somewhere around the late 1980s. So, we were particularly surprised to learn that the Doobie Brothers eventually got back together. What hasn’t surprised us is the ongoing application of artificial intelligence in the creation, dissemination, and aggregation of news. Algorithms are now employed to write news stories with increasing sophistication, while other machine learning platforms work to detect fake news. And many news aggregation platforms such as Google News use AI to automate the search for the storylines it believes will interest each user. Tokyo-based startup SmartNews is an upstart in the AI-powered news aggregation space that just reached a $1.1 billion valuation this month following its latest Series E funding round.
We call private companies that reach the $1 billion threshold unicorns, which once reflected the rare nature of such valuations just a few years ago. Today, unicorns are falling from the sky. In the first half of 2019, 62 new startups joined the unicorn club, according to Pitchbook. The CB Insights unicorn tracker now lists nearly 400 startups at a combined value of $1.1 trillion.
So our interest in SmartNews isn’t just predicated on its shiny new valuation, which comes after raising a relatively modest $116 million since it was founded in 2012. What particularly piqued our curiosity is how this relatively unknown Japanese startup uses AI to aggregate news – and how quickly it has grown in the United States.
AI for News Aggregation and Recommendation
Let’s start with the tech. As we’ve already alluded: AI tools for aggregating news stories into a customizable feed is nothing new. Last year, Google rebuilt its news recommendation platform using a “new set of AI techniques to take a constant flow of information as it hits the web, analyze it in real time and organize it into storylines,” according to a blog post. It claims this approach “understands the people, places and things involved in a story as it evolves, and connects how they relate to one another.” Who needs a news anchor anymore? A survey of nearly 200 publishers by the Reuters Institute for the Study of Journalism published last year found that nearly 60% of digital news platforms use some form of AI for content recommendations.
Credit: Reuters Institute for the Study of Journalism
One of the more popular news aggregators outside of Google is Silicon Valley-based Flipboard, which has raised $210.5 million since it was founded in 2010. It also uses AI for curating the news feed of its user base, which reportedly numbered 145 million monthly users, as of last year. It also claims 11,000 publishers on its platform. However, AI doesn’t do all of the work: A small group of humans is still in the loop, manually selecting articles to recommend to its readers. And let’s not forget about the world’s largest startup, ByteDance, which employs AI to engage its one billion users in China on the Toutiao news platform by “analyzing the features of content, users and users’ interaction with content.” The customized news feeds keep users engaged an average of 74 minutes per day.
On the other hand, SmartNews relies heavily on its algorithms to “evaluate millions of articles, social signals and human interactions to deliver the top 0.01% of stories that matter most, right now.” While the platform does make recommendations based on its users previous activity, it emphasizes what it calls “personalized discovery,” meaning that machine will post content that you probably would have never come across in your normal news bubble.
SmartNews Seeing Success
Apparently, the strategy is working. While its user base is relatively small compared to Google or even Flipboard – there are about 10 to 20 million active users (depending on what stats you believe), mainly between Japan and the United States – SmartNews is growing more rapidly than any other news aggregator. Traffic on the site jumped 145% in 2018, with an average monthly rate of 9%, according to Parse.ly, a data analytics company for audience engagement.
The nearly 400 U.S. publishers and 2,000 Japanese publishers on the platform are fans of the site, as they can expect referral traffic from SmartNews to give them a bump as high as 17% in any given month, Parse.ly reported. More recent data do show that referrals have plateaued a bit this year:
Another feature of SmartNews is that it sends readers to the website of the original article, and allows its partners to insert their own advertising on the news feed while keeping 100% of the revenue, according to an analysis by the Nieman Journalism Lab at Harvard University. In Japan, SmartNews also has a program that drives not just traffic but revenue to its partners, a model that it expects to export to the United States within the next year.
The SmartNews Bump
The SmartNew bump is real. For instance, MIT Technology Review reported nearly 17% growth, month over month, coming from SmartNews, since it became a partner in February, according to the Nieman Lab article. SmartNews now makes up about 4.3% of the MIT tech news site’s overall referral traffic.
In another case study, SmartNews boosted page views for E! News around the annual Met Gala. SmartNews created a featured content set in the E! News channel and scheduled relevant in-app promotions around the event. Working with E! News, SmartNews used specific keywords in the metadata of the entertainment publisher’s content feed to identify and pull the Met Gala stories together under a custom header in its publisher channel. The content, promoted through SmartNews’ Top tab feature, helped give E! News a big bump over the two days of coverage:
We were disappointed to learn that the Met Gala has nothing to do with baseball; it’s some freaky costume show at a museum.
All of this begs the question: How does SmartNews make money? Its business model seems predicated on the idea that if it makes its partners money, then it can make money, too. It’s an old-fashioned idea using 21st century technology. It certainly flies in the face of the current model of most AI news aggregators, especially companies like Facebook that have remained aloof to the concerns of publishers over how the social media site makes money off of other companies’ content, though that appears to be changing. We suspect much of the revenue, at least from the United States, is coming through promotional services such as in the E! News example. SmartNews also lists about 20 featured U.S. publishers, including the Associated Press, The Atlantic, and USA Today, which are probably paying some premium to get top billing.
If SmartNews can keep the momentum going, it may get top billing among the crowded field of news aggregators. One wonders how much longer a startup like SmartNews will escape the
clutches notice of Japan’s SoftBank and its multi-billion-dollar bankrolled Vision Fund, which is investing heavily in AI and related technologies.
Worried that AI might steal your job? You should be. Why not get a Masters in Machine Learning from London’s Imperial College, one of the top-10 universities in the world? And you can study online!